INDEX

Explanations

mentions of news events or criminal activities involving various individuals

oai_token-act-pair · gpt-3.5-turbo

references to significant events or actions involving legal cases or judicial matters

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 yourselves

-0.73

 ourselves

-0.66

 wise

-0.61

 however

-0.61

 sane

-0.60

:-

-0.59

 myself

-0.58

(~

-0.58

 oneself

-0.57

:(

-0.56

POSITIVE LOGITS

aepernick

0.67

 Corinth

0.62

Aren

0.60

interstitial

0.60

ricane

0.59

abortion

0.58

 unfairly

0.57

 illegally

0.57

Islam

0.57

illegal

0.56

Activations Density 1.138%

mentions of news events or criminal activities involving various individuals

references to significant events or actions involving legal cases or judicial matters

No Comments

No Known Activations